Tutorial

Healthc Inform Res. 2017 October;23(4):349-354.

https://doi.org/10.4258/hir.2017.23.4.349
pISSN 2093-3681 • eISSN 2093-369X

Interactive Visualization of Healthcare Data Using

Tableau
Inseok Ko, MS1, Hyejung Chang, PhD2
1
Division of Biomedical Informatics, Seoul National University College of Medicine, Seoul, Korea; 2School of Management, Kyung Hee University, Seoul, Korea

Objectives: Big data analysis is receiving increasing attention in many industries, including healthcare. Visualization plays an
important role not only in intuitively showing the results of data analysis but also in the whole process of collecting, clean-
ing, analyzing, and sharing data. This paper presents a procedure for the interactive visualization and analysis of healthcare
data using Tableau as a business intelligence tool. Methods: Starting with installation of the Tableau Desktop Personal ver-
sion 10.3, this paper describes the process of understanding and visualizing healthcare data using an example. The example
data of colon cancer patients were obtained from health insurance claims in years 2012 and 2013, provided by the Health
Insurance Review and Assessment Service. Results: To explore the visualization of healthcare data using Tableau for begin-
ners, this paper describes the creation of a simple view for the average length of stay of colon cancer patients. Since Tableau
provides various visualizations and customizations, the level of analysis can be increased with small multiples, view filtering,
mark cards, and Tableau charts. Conclusions: Tableau is a software that can help users explore and understand their data by
creating interactive visualizations. The software has the advantages that it can be used in conjunction with almost any data-
base, and it is easy to use by dragging and dropping to create an interactive visualization expressing the desired format.

Keywords: Data Display, Statistics as Topic, Information Storage and Retrieval, Artificial Intelligence

I. Introduction ing an integrated view and evidence for making healthcare

decisions [1,2]. Tableau connects users with a variety of
Data visualization is an intuitive way for users to easily read data sources and enables them to create data visualizations
and understand data, especially in big data analyses. It helps by making charts, maps, dashboards, and stories through a
to improve the quality of policies or services by present- simple drag and drop interface. Although Tableau has re-
ceived some attention in the healthcare field [3], Tableau is
Submitted: September 27, 2017
Revised: October 19, 2017 still not widely used in the healthcare industry. Therefore,
Accepted: October 20, 2017 this paper introduces Tableau and presents the procedure of
using Tableau for the interactive visualization and analysis of
Corresponding Author
healthcare data to encourage its widespread use.
Hyejung Chang, PhD
School of Management, Kyung Hee University, 26 Kyungheedae-ro,
Dongdaemun-gu, Seoul 02447, Korea. Tel:+82-2-961-9432, E-mail:
[email protected]
II. Methods
This is an Open Access article distributed under the terms of the Creative Com- 1. Installing Tableau
mons Attribution Non-Commercial License (http://creativecommons.org/licenses/by- Tableau provides three versions of Tableau Desktop [4]. The
nc/4.0/) which permits unrestricted non-commercial use, distribution, and reproduc-
tion in any medium, provided the original work is properly cited. Public version is free but the Personal and Professional ver-
ⓒ 2017 The Korean Society of Medical Informatics sions are available with a fee or free 14-day trial. However,
Tableau provides a 1-year free product key with support for
Inseok Ko and Hyejung Chang

students and instructors in academic programs. This tutorial product key by sending roof papers by email according to
reflects the Tableau Desktop Personal version 10.3 (64-bit). the procedure shown in Tableau.
To install Tableau, an e-mail address is required for the user
to set up an account [5]. Downloads can be accessed through 2. Connecting to Sample Data
the addresses shown in Figure 1. The user enters the e-mail The sample in this example procedure description includes
address and clicks the ‘Download’ button. Then a download the height and weight data of colon cancer patients from
will be begin automatically. The 32-bit and Mac versions can health insurance claims during the years 2012 and 2013,
be downloaded via the link “Need a different version?” If the provided by the Health Insurance Review and Assessment
user uses 32-bit OS, it can be downloaded with 32-bit ver- Service (HIRA) [7]. Once the data is ready, various files (Ex-
sion automatically. cel, text, Access, and so on) and a server (MS SQL, MySQL,
When the user runs a downloaded file, user agreement to etc.) can be connected via the Connections section to the left
the terms and conditions is required before installation. The of the Tableau window.
user must confirm that agreement is checked and click the This sample is Excel formatted data, so Excel is selected
installation button. Installation will proceed and Tableau will to import the data. The attached datasets appear in the up-
be run automatically. When a Tableau activation window per left corner, and the sheets are listed at the bottom of the
pops up, the user can click a trial or product key buttons. datasets. These consist of 2012 and 2013 sheets. If the 2012
This procedure description is for a trial of Tableau Desktop. sheet is selected and dragged to the ‘Drag sheets here’ sec-
To download a free product, Tableau Public, the user can tion, the 2012 sheet will be added to the section, and the data
click the ‘Download’ button of the Public version [4]. The will be displayed below. To merge the 2012 and 2013 sheets,
download and installation process is the same as that for the the 2013 sheet is dragged to ‘Union’ as shown on the 2012
Tableau Personal version. sheet. Then, the worksheet is clicked next to the bottom data
To obtain a 1-year free product key, the user can apply for source to create a new worksheet. The Tableau workspace
the Academic Programs supported by Tableau [6]. Proce- consists of the following items and functions (see Figure 2):
dures can be followed by accessing the pertinent points (i.e., • Menus: files, data, worksheets, dashboards, stories, etc.
students, instructors, and administration). For example, a • ‌Toolbar: undo, redo, swap rows and columns, show me,
user who has access to students should click on the ‘GET etc.
TABLEAU FOR FREE’ button to enter his or her user infor- • ‌Data Pane: Tableau assigns data fields of the connected
mation on the entry screen. One can receive a 1-year free dataset in the left section. ‘Formatting’ and ‘Analytics’

Figure 1. Downloading a trial of Tableau Desktop (https://www.tableau.com/products/desktop/download).

350 www.e-hir.org https://doi.org/10.4258/hir.2017.23.4.349

 Interactive Visualization Using Tableau

Menus

Toolbar
Shelves and cards

View

Data pane

Figure 2. ‌Interface of workspace in

Sheet tabs
Tableau Desktop.

panes are also available. changed to a total sum, it is possible to change it to different
- ‘Dimensions’
‌ in the upper section include discrete cat- formats, such as average. The total LOS of male patients was
egorical data such as type or date. larger than that of female patients, but there was no differ-
- ‘Measures’
‌ in the lower section contain continuous data. ence between the average LOSs of male and female patients
• ‌Shelves and Cards: page shelf, filter shelf, mark cards, and (Figure 3).
top space with row and column shelves
• ‌View: The right space to create a view (table, graph) by 2. Increasing the Level of Details with Small Multiples
adding fields To increase the level of detail, an additional field, such as
• ‌Sheet Tabs: Create or move to created worksheets, dash- cancer stage, can be added to the previous basic view. Then,
boards, and stories the LOSs can be compared by gender and final stage. The ‘fi-
nal stage’ field of the Dimensions is dragged to the rows shelf
III. Results and placed in front of the LOS field. The same results can be
obtained by dragging the final stage field to the left average
1. Basic View LOS axis of the view.
To start with a basic view for comparing the lengths of stay
(LOSs) of male and female colon cancer patients, ‘gender’ in 3. Filtering the View and Using the Mark Cards
the Dimensions is dragged and dropped into the ‘columns’ After creating a basic view using rows and columns shelves,
shelf. The gender field is displayed in blue. Then, “LOS” in mark cards and filters can be used to create a higher level of
the Measures is dragged to the ‘rows’ shelf. The LOS field views. Explicitly, the view can be filtered to focus explora-
is displayed in green. The continuous field is automatically tion, and mark cards can be used to add depth to the analy-
calculated as a total ‘sum’ and appears as a bar graph. This sis. To compare the LOSs by gender and final stage after fil-
shows that the total LOS of male patients was larger than tering C180, C19, and C20 of the ‘main code’, the main code
that of female patients. By selecting the LOS field that has field of the Dimensions is dragged to the ‘filter’ shelf. The fil-

Vol. 23 • No. 4 • October 2017 www.e-hir.org 351

Inseok Ko and Hyejung Chang

Figure 3. Getting started with a basic view.

Figure 4. Example of filtering views and using mark cards.

ter [Main code] pop-up window appears. In the ‘general’ tab, is pressed. The view is updated with filtered data. Next, the
C180, C19, and C20 are checked, and then the ‘OK’ button main code field is dragged into the ‘color’ of the mark cards.

352 www.e-hir.org https://doi.org/10.4258/hir.2017.23.4.349

 Interactive Visualization Using Tableau

The filtered main code is updated to indicate the gender and the Measures is dragged to the color of the mark cards, and
the average LOS according to the final stage combined with the sum is changed to the average of the weight field. Then,
the generation of the color legend (Figure 4). the color can be edited the color. ‘Red-Blue-White Diverg-
ing’ is selected, ‘Reversed’ is checked, and then the OK but-
4. Building a Tableau Chart (Treemap) ton is clicked. The average LOS determines the size of the
Although the stacked bar chart is created automatically, rectangle, and the average weight determines the color of the
it is possible for Tableau to change various graphs in the rectangle (Figure 5).
‘Show Me’ section [8]. For example, to change the graph to
‘Treemap’ on the toolbar, one should click Show Me in the IV. Discussion
upper-right corner, and then select Treemap. It appears as
a changed tree map automatically. The size and color of the By following the procedure described in this paper, it is pos-
rectangle in the Treemap is determined according to the av- sible to create a simple view to explore sample data using
erage LOS. The longer the average LOS, the darker the color Tableau for beginners. Tableau provides various visualiza-
and the larger the size of the rectangle. The main code field tions, such as tables, heatmaps, maps, bar charts, treemaps,
is dragged and dropped to the color of the mark cards. De- line charts, bubble charts, and so on as well as customiza-
pending on the values of the previously filtered main code tions, such as radar charts with user intention. Users can
field, the color of the rectangle changes. The ‘weight’ field in also create interactive visualizations that express the desired

Figure 5. Example of a Tableau chart, Treemap.

Vol. 23 • No. 4 • October 2017 www.e-hir.org 353

Inseok Ko and Hyejung Chang

format using filters, drop-down lists, the calculated field, and 3. Martinez R, Ordunez P, Soliz PN, Ballesteros MF. Data
so on. This procedure is intended for beginner level, so in- visualisation in surveillance for injury prevention and
depth functions and detailed options for intermediate and control: conceptual bases and case studies. Inj Prev
advanced users will be covered in the forthcoming paper. 2016;22(Suppl 1):i27-33.
4. Tableau. Meet the Tableau Desktop family [Internet].
Conflict of Interest Seattle (WA): Tableau; c2017 [cited at 2017 Sep 2].
Available from: https://public.tableau.com/en-us/s/
No potential conflict of interest relevant to this article was download.
reported. 5. Tableau. Downloading a trial of Tableau Desktop [Inter-
net]. Seattle (WA): Tableau; c2017 [cited at 2017 Sep 2].
Acknowledgments Available from: https://www.tableau.com/products/desk-
top/download.
This work was supported by the National Research Founda- 6. Tableau. Academic programs [Internet]. Seattle (WA):
tion of Korea (NRF) grant funded by the Korea government Tableau; c2017 [cited at 2017 Sep 2]. Avaiable from: https://
(No. 2015S1A5A2A01014390). www.tableau.com/academic.
7. Healthcare Bigdata Hub. Public data of colorectal can-
References cer patients in 2012 and 2013 [Internet] Wonju, Korea:
Health Insurance Review and Assessment Service; c2017
1. Wang L, Wang G, Alexander CA. Big data and visual- [cited at 2017 Sep 2]. Available from: http://opendata.
ization: methods, challenges and technology progress. hira.or.kr/op/opc/selectOpenData.do?sno=1&publData
Digit Technol 2015;1(1):33-8. TpCd=&searchCnd=&searchWrd=&pageIndex=4.
2. Janvrin DJ, Raschke RL, Dilla WN. Making sense of 8. Mackinlay J, Hanrahan P, Stolte C. Show me: automatic
complex data using interactive data visualization. J Ac- presentation for visual analysis. IEEE Trans Vis Comput
count Educ 2014;32(4):31-48. Graph 2007;13(6):1137-44.

354 www.e-hir.org https://doi.org/10.4258/hir.2017.23.4.349